Early Land-Use and Landscape Development in Arisaig

Early Land-use and Landscape Development in Arisaig

by Stephen Carter, 1 Magnar Dalland1 and Deborah Long2

1Headland Archaeology Ltd, 13 Jane Street, Edinburgh EH6 5HE 2Plantlife Scotland, Balallan House, Allan Park, Stirling FK8 2QG

with a contribution by Caroline Wickham-Jones

Scottish Archaeological Internet Report 15, 2005 www.sair.org.uk Published by the Society of Antiquaries of Scotland, www.socantscot.org with Historic Scotland, www.historic-scotland.gov.uk and the Council for British Archaeology, www.britarch.ac.uk

Editor Debra Barrie

Produced by Archetype Information Technology Ltd, www.archetype-it.com

ISBN: 0 903903 84 9 ISSN: 1473-3803

Requests for permission to reproduce material from a SAIR report should be sent to the Director of the Soci- ety of Antiquaries of Scotland, as well as to the author, illustrator, photographer or other copyright holder. Copyright in any of the Scottish Archaeological Internet Reports series rests with the SAIR Consortium and the individual authors. The maps are reproduced from Ordnance Survey material with the permission of Ordnance Survey on behalf of The Controller of Her Majesty’s Stationery Office. ©Crown copyright 2001. Any unauthorized reproduction infringes Crown copyright and may lead to prosecution or civil proceedings. Historic Scotland Licence No GD 03032G, 2002. The consent does not extend to copying for general distribution, advertising or promotional purposes, the creation of new collective works or resale. Contents

1 Summary ...... 1

2 Introduction...... 2

3 The Archaeology of Arisaig: Early Investigations ...... 4

4 Absence or Invisibility? ...... 5

5 A Transect Across the Landscape: the Line of the A830 Improvements ...... 6

6 The Archaeological Investigations ...... 8 6.1 Shieling huts and circular stone features: Sites 3–6 (centred on NM 6675 9005)...... 8 6.1.1 Survey ...... 8 6.1.2 Evaluation ...... 8 6.1.3 Excavation ...... 8 6.1.4 Reuse of temporary sites from prehistory ...... 13 6.2 Shieling huts and cultivation rigs: Site 8 (centred on NM 659 888)...... 13 6.2.1 Survey ...... 13 6.2.2 Evaluation ...... 13 6.2.3 Excavation ...... 14 6.2.4 Lithics by C Wickham-Jones...... 17 6.2.5 Reuse and expansion over time ...... 17 6.3 Rectangular turf structure: Site 10 (NM 6583 8835)...... 17 6.3.1 Survey ...... 17 6.3.2 Evaluation ...... 18 6.4 Township: Site 15 (centred on NM 658 878) ...... 18 6.4.1 Survey ...... 18 6.4.2 Evaluation ...... 18 6.4.3 Watching brief ...... 18 6.5 Rectangular building: Site 26 (NM 6599 8693) ...... 18 6.5.1 Survey ...... 18 6.5.2 Evaluation ...... 18 6.6 Kerb cairn: Site 41 (NM 6634 8651) ...... 18 6.6.1 Excavation ...... 19 6.6.2 History and use ...... 22

7 Vegetation History ...... 25 7.1 Peat core from Allt Dail an Dubh-asaidh ...... 25 7.2 Other evidence for vegetation history of the Arisaig area ...... 26

8 Occupation and Use of a Marginal Landscape...... 27 8.1 Core and periphery in Arisaig ...... 27 8.2 Site-based and landscape-scale evidence for human impact...... 27 8.3 Evolution of the Arisaig landscape ...... 29 8.3.1 Stage 1: Early prehistory ...... 29 8.3.2 Stage 2: Later prehistory ...... 30

iii 8.3.3 Stage 3: Medieval and later period ...... 30 8.3.4 Postscript: The recent past ...... 30

9 Palynological Analyses at Allt Dail an Dubh-asaidh ...... 32 9.1 Introduction and summary ...... 32 9.2 Methodology ...... 32 9.3 Radiocarbon dates ...... 32 9.4 Results...... 32 9.5 Interpretation of pollen data ...... 34 9.5.1 Zone 1A: Base to c 8900 BP ...... 34 9.5.2 Zone 1B: c 8900–7245 BP ...... 35 9.5.3 Zone 1C: c 7245–6070 BP ...... 35 9.5.4 Zone 1D: c 6070–4520 BP ...... 41 9.5.5 Zone 1E: c 4520–2410 BP ...... 42 9.5.6 Zone 1F: c 2410–present ...... 42 9.6 Interpretation of charcoal data ...... 43 9.7 Natural vegetation succession...... 43 9.8 Human activity...... 43 9.9 Conclusions ...... 43

10 Acknowledgements ...... 45

11 References ...... 46

iv List of illustrations

1 Site location (from Ordnance Survey maps © Crown copyright) ...... 2 2 Detailed map showing location of pollen sites and known medieval and prehistoric sites (from Ordnance Survey maps © Crown copyright) ...... 3 3 Map showing site locations along the proposed A830 road placement (from Ordnance Survey maps © Crown copyright) ...... 7 4 Contour map showing the location of Sites 3–6 with evaluation and excavation trenches ...... 9 5 View of Site 6 from the south-west...... 10 6 Plan of Sites 6B and 6D...... 11 7 Plan of the excavated features on Site 6A ...... 12 8 Calibrated radiocarbon determinations from Site 6A...... 13 9 Contour map showing placement of Site 8...... 14 10 View of Site 8 before excavation, from the north...... 14 11 Calibrated radiocarbon determinations from Site 8 ...... 15 12 Plan of Site 8: (a) phase I: basal occupation deposit; (b) phase II: lower structure; (c) phase III: middle structure; (d) phase IV: upper structure...... 16 13 Contour map showing location of Site 41...... 19 14 Site 41: view of cairn after turf removal, from the south-west...... 19 15 Site 41: ring of kerb-stones, viewed from the north ...... 20 16 Site 41: plan of kerb and robbed out cist with basal layer of stones ...... 21 17 Calibrated radiocarbon determinations from Site 41 ...... 22 18 Site 41: cairn after removal of topsoil ...... 23 19 Site 41: sections of the kerb cairn from (a) the north (Profile 1) and (b) the east (Profile 2)...... 24 20 Changes in tree pollen percentages from Polnish, Lochan Doilead and Allt Dail an Dubh-asaidh ...... 28 21 Stages of landscape evolution near Arisaig: (a) early prehistory: 7500–3550 BC; (b) later prehistory: 3550 BC to AD 500; (c) medieval and later period: AD 500–1800 (from Ordnance Survey maps © Crown copyright) ...... 29 22 Sketch map showing core locations ...... 33 23 Long traverse basin depth profiles ...... 34 24 Short traverse basin profiles...... 34 25 Allt Dail an Dubh-asaidh: percentage pollen diagrams for (a) trees and shrubs, (b) herbs and (c) acquatics...... 36 26 Allt Dail an Dubh-asaidh: pollen concentration curves for (a) trees and shrubs and (b) herbs, ferns and mosses ...... 39 27 Allt Dail an Dubh-asaidh: pollen condition classes...... 41

v List of tables

1 Radiocarbon determinations of samples from Site 6A ...... 11 2 Radiocarbon determinations of samples from Site 8...... 15 3 Radiocarbon determinations of samples from Site 41...... 22 4 Pollen zones, identifying main local and extra-local and regional vegetations types...... 25 5 Pollen zones with radiocarbon dates and accumulation rates ...... 33 6 Stratigraphic descriptions of pollen core ...... 35

vi 1 Summary

Re-alignment of a 6-km section of the A830 road in medieval and post-medieval periods. In both cases, Arisaig provided an opportunity to investigate the Bronze Age deposits were also recorded at the base of archaeology of this poorly understood area of the the medieval sequence. Analysis of a long peat core West Highlands. A combination of archaeological from a basin close to one of the shielings revealed a and palaeoenvironmental techniques were used to history of continuous but gradual decline in investigate the road corridor. Archaeological survey, woodland, starting in about 3200 BC and continuing followed up by selected excavations, identified a to the present day. Collation of archaeological and previously unrecorded Bronze Age kerb cairn and palaeoenvironmental data from the present project two areas of shieling huts. Investigation of the and previous investigations in the area have allowed shielings obtained evidence for repeated reuse of the creation of a tentative model of landscape sites and reconstruction of structures through the evolution for Arisaig.

1 2 Introduction

Arisaig, together with Moidart to the south and logical investigations in recent years (eg Rees 1996), Morar and Knoydart to the north, make up the area in line with current national planning policy traditionally known as na garbh Chriochan: the guidance. We present the results of a programme of Rough Bounds (Illus 1). An anonymous writer in archaeological and palaeoenvironmental investiga- 1750 (quoted by Rixon 2002) considered that: tion carried out during the upgrading of the section of the A830 between the village of Arisaig and All these countries viz. Knoydart, the Two Morrirs, Kinsadel at the west end of Loch Morar (Illus 2). This Moydart, and Arisaig, are the most Rough Moun- 6-km section of road was rebuilt on a totally new line, tainous and impassible parts in all the Highlands inland from the existing road, and therefore offered of Scotland, and are commonly called by the Inhab- an opportunity to record an undisturbed transect of itants of the Neighbouring countries the highlands the landscape. The investigations were carried out of the Highlands. by the Centre for Field Archaeology (CFA), Univer- The construction of the Fort William to Mallaig sity of Edinburgh, and Headland Archaeology Ltd Road (the ‘Road to the Isles’, completed in 1812) and between 2000 and 2001; the road was opened in the railway that follows the same route (opened in 2003. 1901) considerably improved the accessibility of For readers of this paper who know the area well, it Arisaig and Morar. Knoydart remains one of the should be noted that we use the name Arisaig to refer least accessible parts of the Scottish mainland, at to coastal land between Loch nan Uamh (Borrodale) least if we judge by the standards of our modern in the south and the River Morar in the north. This car-dependent society. definition does not precisely match the historic Progressive upgrading of the road (now the A830) estate of Arisaig as it takes in part of South Morar in from its former single-track status to modern trunk the north and excludes parts of Arisaig to the road standards has been accompanied by archaeo- south-east.

Loch Hourn

70 80 Ladhar Bheinn Arisaig Roinn na Beinne 1020

Sound of Sleat 441 K N O Y D A R T 100

Loch Nevis Mallaig Sgurr na Sgurr Ciche Bhuidhe 1040 440

NORTH MORAR Loch Morar 90 Loch Morar Sgur Thuilm Arisaig SOUTH MORAR Sidhean Mor 963 599

Illus 02 A8A83083030 Loch nan Uamh Railway 80 Loch Ailort Sound of Arisaig Beinn Odhar Bheag 882

M O I D A R T

Loch Shiel N Meall Mor Loch Shiel 759

0 10 km Land over 200m

Illus 1 Site location (from Ordnance Survey maps © Crown copyright)

2 795000

8 Sound of Sleat

Loch Morar

Allt Cam Carach 2 Site 6A 790000

Site 8 3 6

A r i s a i g

7 Loch nan Ceall Site 41 9 4 10 11 Railway 12 Brunery burn 13 785000 A830 14 15 16 5

Loch nan Uamh N

17 0 1km C Crown copyright. 165000 170000

KEY Pollen sites Pollen site 1: Lochan Doilhead (Williams 1977) 2: Allt Dail an Dubh-asaidh 3: Mointeach Mhor (Shennan et al 1994) 9: NMRS NM68NE 7: Kerbed cairn Medieval site 4: Loch nan Eala (Shennan et al 1995) 10: NMRS NM68NE 2: Crannog Prehistoric site 5: Polnish (Clarke 2000) 11: NMRS NM68NE 4: Oak platform 12: NMRS NM68NE 6: Cairn Archaeological sites 13: NMRS NM68NW 9: Promontory fort Limit of study 14: NMRS NM68SE 1: Cup-marked stone Land over 200 m 6: NMRS NM68NW 1: Stronghold and Corn-Drying Kiln. Medieval or later 15: NMRS NM68SE 5: Cairn 7: NMRS NM68NE 1.00: Medieval church 16: NMRS NM68SE 2: Promontory fort 8: NMRS NM69SE 1: Mixed scatter. Quartz and flint tools. Mezolithic - Bronze Age 17: NMRS NM68SW 2: Fort

3 3 The Archaeology of Arisaig: Early Investigations

This beautiful but remote part of the west coast of best known for his investigations of archaeological Scotland has received little attention from archaeolo- sites in Kilmartin, Argyll. Blundell was one of the gists and antiquarians in the past. In part this must pioneers in crannog research in Scotland; he carried reflect an absence of major sites that might attract the out limited excavations in south-west Scotland and attention of distant researchers: there are no visually the Highlands. Later, the discovery of flint and quartz impressive sites in Arisaig and prehistory is either tools prompted AD Lacaille to investigate the site in largely invisible or absent. However, the few discov- person. Despite these and other records, our eries that have been made did attract attention from knowledge of the archaeology of Arisaig is insubstan- contemporary investigators. A crannog revealed by tial and readily summarized in Illus 2.Knownsites the draining of Loch nan Eala was subsequently are concentrated in the vicinity of the present village, visited by Reginald Mapleton in 1868 and Odo there is only one site pre-dating the Bronze Age and Blundell in 1911. Mapleton was the Episcopalian the medieval period is represented solely by the late minister to the Malcolm family of Poltalloch and is medieval church at Kilmory.

Illus 2 (opposite) Detailed map showing location of pollen sites and known medieval and prehistoric sites (from Ordnance Survey maps © Crown copyright)

4 4 Absence or Invisibility?

The modest nature of the archaeological record for masking prehistoric houses and enclosures, as has Arisaig inevitably raises the question of whether this been recognized elsewhere along the west coast of reflects a genuine scarcity or simply a problem of Scotland (Ritchie et al 1974). It is also supported by invisibility. Rixon, attempting to describe the recent high spatial resolution palynological studies in prehistoric period in Arisaig and Morar, deals with West Affric that have demonstrated the presence of this question in the following terms: prehistoric farming communities in upland landscapes where no surface archaeological evidence for There are two ways to look at the prehistoric period their former presence has been detected (Davies & in Arisaig and Morar: either as an almost blank Tipping 2005). This finding requires us to accept that chapter in which we can do no more than list those the archaeology of the uplands (or marginal lowland prehistoric monuments which have been discov- areas) need not be visible, despite the presence of ered to date; or as an unopened treasure-chest for extensive tracts of upstanding prehistoric archae- the future (Rixon 2002). ology in the Highlands of Scotland, or even detectable Arguments in favour of genuine absence include by conventional archaeological techniques of field the fact that the area is dominated by unimproved survey and excavation. land where the destruction and levelling of archaeo- Weighing up the opposing arguments, it seems logical sites is unlikely to occur: sites, if present, likely that neither is wholly correct. The landscape of should be visible. The area has been examined by Arisaig is not uniform and must have offered varying archaeologists, although not subject to recent potential for settlement and use by earlier inhabit- systematic archaeological survey (as similar areas in ants. The clustering of recorded archaeological sites Argyll have, for example by RCAHMS), and it seems correlates with what is currently the better agricul- unlikely that any additional major upstanding tural land so it seems reasonable to propose that monuments exist. Absence of archaeological sites there is a real focus of past settlement around the also matches our perception of this area (the ‘Rough current village of Arisaig. Moving out from this core Bounds’) as marginal and capable only of supporting area into what is currently bog or very rocky grazing a small human population. land it is reasonable to propose that the absence of The counter argument for invisibility can draw on archaeological sites is real. the presence of widespread peat deposits, potentially

5 5 A Transect Across the Landscape: the Line of the A830 Improvements

The position of the new road line provided an oppor- weaknesses inherent in a project of this type. On the tunity to test the idea of a core settlement area in one hand, the investigation of a road corridor Arisaig surrounded by less-favoured land. The road provides an archaeologically random sample of the line starts within the proposed core at its south end landscape. It forces the archaeologist to examine and then runs northwards out into an apparently ground that would otherwise be passed over on the ‘blank’ area where the archaeological potential way to a more promising destination. In these might be predicted to be low. The project employed circumstances, the potential for serendipitous two approaches in the study of this landscape: discoveries is high as the archaeologist is less able to site-based archaeological investigations of the allow preconceptions to drive the design of the physical remains of past human activity and palaeo- project. On the other hand, a road corridor is unlikely environmental investigations of peat deposits to to provide a structured sample of a landscape and understand human impact on a landscape scale. It will not allow the archaeologist to examine the full was hoped that the two approaches would provide range of habitats in an area. In the case of Arisaig, complementary information on how the landscape the road line runs parallel to and inland from the had been utilized and evolved through time. coast. As a result, it avoids both the narrow culti- It must be stressed that the nature of the project vated coastal zone dominated by windblown sand was necessarily opportunistic as the position of the and the higher hills that lie inland. We therefore sample transect across the landscape was deter- collected no new data on these habitats and this has mined entirely by the position of the new road. limited our ability to understand the evolution of the Consequently, it displays both the strengths and the landscape as a whole.

Illus 3 (opposite) Map showing site locations along the proposed A830 road placement (from Ordnance Survey maps © Crown copyright)

6 KEY Recorded archaeological sites N N SitesPrehistoric selected for evaluation 8 LandForested over 200 m Gazeteer0 1km 810000

Sound of Sleat

63 64 65

Proposed road Illus 4 Sites 3+6

790000

Road corridor

Illus 9 Site 8

A830

Allt Cam Carach

Site 10

Railway

Site 1015

A r i s a i g

Site 26

Brunery burn Illus13 Site 41

Loch nan Ceall C Crown copyright. 165000

7 6 The Archaeological Investigations

The programme of archaeological works took a was reused as a sheepfold. To the north of the typical staged approach and was designed to test and building there was an area of cultivation rigs that then investigate the archaeological potential of the was probably contemporary with the building. The route. The initial stages included a desk-based study cultivation remains covered an area 30m by 90m and followed by a walkover survey of the road line. This comprised rigs 1.5–2m wide. By 1873, they had been identified 44 visible features; the vast majority were abandoned and a stone dyke (Site 4) had been built clearly structures of 19th and 20th century date and across the rigs. these were concentrated at the south end of the route within the proposed core area (Illus 3). The potential for invisible sub-surface archaeological features was 6.1.2 Evaluation tested in two ways. Engineers’ test pits were monitored at intervals of 75m and the southern 3km Four structures were investigated further. Evalua- of the route was monitored during topsoil stripping. tion trenches were cut through examples of shieling Nothing of interest was noted in the test pits and huts (Sites 3A and 6C) and two of the circular stone only one undated pit was identified during the features (Sites 6A and B). topsoil strip. The evaluation trenches across the two shieling Seven of the sites identified during the survey huts exposed the footings of turf and stone walls. No were selected for further evaluation as they were traces of any occupation deposits were seen above either of potentially early date (pre-19th century) or the natural subsoil nor were any artefacts retrieved. might conceal earlier features (Illus 3). Three of The size and shape of the structures indicated that these sites were then subject to more substantial they represented remains of shieling huts and the excavation on the basis of the evaluation results. absence of artefacts suggested an 18th-century or earlier date (19th-century structures elsewhere along the road line contained abundant pottery, 6.1 Shieling huts and circular stone glass and metalwork). features: Sites 3–6 (centred on The two circular stone features (Sites 6A and 6B) NM 6675 9005) were found to be sitting on and within the upper layerofturfandtopsoil(Illus 6). No artefacts were 6.1.1 Survey retrieved and their date and purpose remained uncertain. The two examples of rings of stones A group of sites was recorded on a south-east facing (Sites 6B and 6D) are similar to features recorded hillside beside a small stream, the Allt Dail an during a survey of the Tangaval peninsula by Dubh-asaidh, on the north side of the Mointeach University of Sheffield (Branigan & Foster 1995). Mhór (Illus 4). The hillside contains pockets of Here they have been interpreted as weights to hold deeper soil, now largely covered in bracken, in down a temporary tent-like shelter for a herdsman. amongst extensive rocky outcrops partly covered in This interpretation could not apply to the third site heather (Illus 5). There are more extensive areas of (6A), which is a circular area of flat stones. An alter- deeper soil along the stream, also bracken-covered. native interpretation for this, and the other two The structures were located in these areas of deeper rings, is that they are bases for peat stacks – fuel for soil; they included a small enclosure, the footings of the occupants of the shieling huts or the farmstead. six small sub-rectangular buildings (Sites 3A–E and During the excavation of the trench across Site 6A, 6C) identified as shieling huts, and three circular a thin lens of charcoal was encountered, 0.25–0.35m stone features 2–3m in diameter and of uncertain below the ground surface; the charcoal deposit was function (Sites 6A, 6B and 6D). The shieling huts apparently not associated with the stone structure measured up to 3m by 4m and were defined by low seen on the surface. A sample was taken from the turf and stone banks spread to 1.5–2m wide and up to deposit for radiocarbon dating (AA-41068/GU-9251) 0.4m high. and produced a 15th century AD date and, on the The ruin of a 19th-century building (Site 5) was basis of this result, an excavation was carried out to situated on level ground between Sites 3 and 6. It further investigate its context. was 3m wide by 16m long and subdivided into three rooms. This structure is marked as a sheepfold on the First Edition OS map surveyed in 1873. However, 6.1.3 Excavation (Illus 7) the walls at the north end were up to 2m high and built from well-fitted faced stones. It appears that The deposits within the excavated area were 0.4– the structure was a building (farmstead?) that later 0.6m deep down to bedrock and natural sand

8 000

100

50 m

m0 4

Rig & Furrow

rroc da

Recorded archaeological site oR

Site 3A

800

Site 3E

a r

nru nru

b b m03

Site 5

nr nr

m u b 0 b

5 60m

40 m 40 da

700 daoR desopo daoR oR

de 1

166700 sop

Dyke

rP orP

Illus 6

Site 4: Dyke

Site 6

3 Illus 4 Contour map showing the location of Sites 3–6 with evaluation and excavation trenches

rod

i rroc d rroc

a oR

600

m m

04 04

m m

05 05

9 Illus 5 View of Site 6 from the south-west

(context 663). In the north-eastern part of the trench To the west of the pit, also overlaying context 650, there was a thin deposit of patchy black sandy silt was a linear stone feature (context 655), running with abundant charcoal (context 662) on top of roughly north to south. It comprised a row of flat bedrock and sealed by a layer of black friable peat stones 0.2–0.5m across. The feature petered out underlying a deposit of sandy silt. A shallow oval pit towards the southern edge of the trench, but con- or scoop (context 653) was cut into the sandy silt tinued northwards into an area of natural tumbled layer. The cut of this pit was not clear and its outline stones and boulders where it seemed to end. The was best defined by its basal deposit of charcoal in a exposed part of the feature measured some 6m matrix of dark brown silty clay (context 654). north–south. It was covered by turf and topsoil and The pit was sealed by a deposit of mid yellowish- underlay the surface circular stone feature (Site brown sandy silt (context 657) with sparse sub- 6A). angular cobbles and rare charcoal flecks throughout. A further three radiocarbon samples were sub- A few fragments of charred hazelnut shell were mitted for dating. These had been taken from the retrieved from a sample taken from the deposit. The basal charcoal deposit (context 662), the fill layer underlay a spread of dark brownish-grey silty (context 654) of the lower pit (context 653) and clay with charcoal flecks (context 650). This deposit from the spread (context 650). This spread was cut was 0.09m deep and filled a shallow depression. No by the upper pit (context 651/608), from which the artefacts were retrieved from the layer, but small radiocarbon sample had been taken during the fragments of burnt bone and hazelnut shell were evaluation. The results are presented in Table 1 retrieved from a sample. The sample also produced and Illus 8. a small quantity of oat grains. Context 650 was cut The date from the basal deposit (context 662) by an oval pit (context 651); the lower fill of this pit indicates that there was activity in the area during (context 608/652) was the charcoal-rich deposit the Bronze Age. However, no features were recorded recorded during the evaluation and dated to the within the trench that could be associated with this 15th century AD. In addition to charcoal, that deposit. The layer of peat that sealed the Bronze Age deposit (context 608/652) also contained a few oat remains must have formed over the site during the grains. The upper fill of the pit comprised light following 2000 years before activity again resumed brown silty sand (context 664). in the early medieval period. The three later dates

10 Table 1 Radiocarbon determinations of samples from Site 6A

Lab code Context Material Lab age BP dC13 1 sigma cal 2 sigma cal SUERC-2441 662 Quercus 3055 ± 40 –25.6‰ 1390–1260 BC 1420–1210 BC (93.0%) (GU-11898) 1200–1190 BC (1.2%) 1140–30 BC (1.2%) SUERC-2443 654 Corylus 1115 ± 40 –23.5‰ AD 890–985 AD 780–800 (1.8%) (GU-11899) AD 810–1020 (93.6%) SUERC-2444 650 Quercus 840 ± 45 –25.7‰ AD 1160–1260 AD 1040–1100 (11.2%) (GU-11900) AD 1110–1290 (84.2%) AA-41068 608/ Quercus 440 ± 35 –26.0‰ AD 1435–65 AD 1418–88 (GU-9251) 652

Site 6D

789990

Site 6A

Evaluation Site 6B trenches

166660

0 5 m

Illus 6 Plan of Sites 6B and 6D

11 Site 6A

657 655

650 662 663 Sondage

789985

651 Evaluation trench 653

657 655

Limit of excavation

166675

02 m

Illus 7 Plan of the excavated features on Site 6A

12 AA-41068 (GU-9251) 440–35BP Context 608/652

SUERC-2444 (GU-119900) 840–45BP Context 650

SUERC-2443 (GU-11899) 111 5–40BP Context 654

SUERC-2441 (GU-11898) 3055–40BP Context 662

2000CalBC 1000CalBC Cal BC/ Cal AD 1000CalAD 2000CalAD

Illus 8 Calibrated radiocarbon determinations from Site 6A indicate that there were episodes of activity through- then the presence of a 19th century farmstead marks a out the medieval period. Apart from the pits, the only significant change of land use from summer grazing to feature that could be associated with this activity permanent settlement. However, this change appears was the linear stone feature (context 655), which to have been short-lived as by the later 19th century could have been contemporary with context 651, the the building had been converted into a sheepfold and late medieval pit. The circular stone features are the area reverted to pasture. This land use has clearly later than the 15th century AD but it is not continued up to the present day. known by how long a period.

6.1.4 Reuse of temporary sites from 6.2 Shieling huts and cultivation prehistory rigs: Site 8 (centred on NM 659 888) The results of the excavation show that the area was used in the middle Bronze Age. The nature of 6.2.1 Survey the remains give us no clues as to the nature of the activity at this time and it is not even certain that The site was located on rocky ground on the west side the dated feature is the product of human activity. of Mointeach Mhór, some 50m to the north of Allt The next evidence of activity in the area was dated Cam Carach. It comprised an extensive complex of to the ninth and tenth centuries AD.Againthe cultivation rig remains covering an area of some evidence is limited, but a series of stratified 100m by 200m (Illus 9). At the north-west of this area features produced dates indicating that the site there were turf and stone footings of a sub- was in continuous use or repeatedly re-visited rectangular shieling hut measuring 8m by 4.5m through the medieval period. The presence of oat overall. A small rectangular annexe 1.5m by 3.5m grains in the medieval deposits suggests that food was situated immediately to the north. A third was prepared and consumed at the site but positive structure was situated at the south-eastern limit of evidence for the reason behind this is not available. the cultivated area. It was defined by a discontin- The deposition of a substantial volume of mineral uous line of stones along the edge of a platform, 2.5m sediment over the post-Bronze Age peat soil in the by 4 m, set up against a vertical rock face (Illus 10). medieval period requires explanation and Of the three structures, only this one was directly in suggests that we may have dug through the highly the line of the new road. dispersed remains of unsubstantial turf and timber structures, essentially a stack of turf shieling huts, spanning the medieval period. The 6.2.2 Evaluation line of stones could then be interpreted as a wall foundation. To investigate this platform further, an evaluation In this case, the adjacent shieling hut (Site 6C) is trench was cut across the feature. A charcoal layer simply the most recent hut to be built on the site and, was exposed at the base of the trench, 0.55m below together with the huts at Site 3, represents the the surface, and was dated to the Bronze Age continued use of a summer shieling in the area into the (AA-41069/GU-9252). On the basis of this date, post-medieval period. If this interpretation is accepted further excavation was carried out.

13 Shieling huts

m02

m5.71

m51

daoR desoporP daoR

788800 rodirroc daoR rodirroc 788800 rodirr Rig & Furrow

c daoR c

12 5.m

10m m5.7 Site 8 Track

Allt Cam Carach

165900 N 788788700700

0 100 m 116565800800

Illus 9 Contour map showing placement of Site 8

6.2.3 Excavation

A 4m by 5m trench was opened up against the foot of the rock face, centred on the evaluation trench. The stratigraphy of the site can be grouped into five structural phases; of these the lower four were dated by radiocarbon (Table 2 and Illus 11).

Phase 1: Basal occupation deposit (Illus 12a)

The basal deposits in the trench comprised two areas of dark brownish-grey silty sand (context 868, with concentrations of charcoal overlying mid yellow to brown sand; context 864, natural subsoil). Context 868 was the same sediment as context 803, which was sampled during the evaluation and dated to between the 11th and 14th centuries BC. One quartz flake was retrieved from this context but there were no features associated with it. A general layer of yellowish-brown silty sand with occasional fragments of charcoal sealed the occupation deposits. It was interpreted as a deposit formed by natural processes, probably hillwash from the slope to the north.

Illus 10 (left) View of Site 8 before excavation, from the north

14 Table 2 Radiocarbon determinations of samples from Site 8

Lab code Context Material Lab age BP dC13 1 sigma cal 2 sigma cal AA-41069 803 = 868 Quercus 2990 ± 45 –26.3‰ 1366–1129 BC 1386–1049 BC (GU-9252) SUERC-2445 865 Pomodieae 380 ± 40 –23.2‰ AD 1440–1530 (48.9%) AD 1440–1530 (55.1%) (GU-11901) AD 1590–1630 (19.3%) AD 1540–1640 (40.3%) SUERC-2446 861 Corylus 245 ± 40 –27.5‰ AD 1530–50 (6.6%) AD 1510–1600 (18.9%) (GU-11902) AD 1630–80 (39.0%) AD 1620–90 (42.6%) AD 1770–1810 (18.9%) AD 1730–1810 (28.2%) AD 1930–50 (3.8%) AD 1920–50 (5.7%) SUERC-2450 859 Corylus 175 ± 40 –27.1‰ AD 1660–90 (12.5%) AD 1650–1890 (79.3%) (GU-11903) AD 1730–1810 (43.0%) AD 1910–60 (16.1%) AD 1920–50 (12.7%)

SUERC-2450 (GU-11903) 175–40BP Context 859

SUERC-2449 (GU-11902) 245–40BP Context 861

SUERC-2445 (GU-11901) 380–40BP Context 865

SUERC-2445 (GU-9252) 2990–45BP Context 803-868

2000CalBC 1000CalBC CalBC/CalAD 1000CalAD 2000CalAD

Illus 11 Calibrated radiocarbon determinations from Site 8

Phase 2: Lower structure (Illus 12b) charcoal sample taken from the occupation deposit. The calibrated date is not very precise due to the On top of the hillwash were the remains of a shape of the calibration curve at this point; the prob- structure that comprised the base of a straight wall ability density is distributed over four distinctive (context 866) set at a 45-degree angle against the peaks between AD 1530 and 1950 at 1 sigma proba- rock face. This wall appeared to be a windbreak bility (Illus 11). However, consideration of the date of forming a shelter open to the south. It enclosed a stratigraphically related samples indicates that it is deposit of dark brownish-grey silty sand with char- not likely to date to the 20th century or the 16th coal and grains of hulled barley (context 865). A century. It is therefore most likely that the date of sample of charcoal taken from context 865 was dated the structure lies within the two middle peaks and to the 15th or 16th century AD. A layer of stony brown dates to the 17th or 18th century AD. silty sand (context 863), thought to be levelled wall material from the shelter wall, covered the wall footing and deposit. Phase 4: Upper structure (Illus 12d)

Immediately on top of the middle structure were the Phase 3: Middle structure (Illus 12c) remains of a third construction of similar shape and style of construction. It was defined by the remains of On top of the levelled wall material were the remains a wall comprising a curving line of angular to of a smaller structure. It was defined by wall footings sub-angular stones (context 852). The structure comprising a curving band of angular to sub-angular occupied the same spot, but enclosed an area of 8m2, stones (context 862), which formed an area 1.5m by twice that of the previous structure. As before, the 2.5m against the rock face with an entrance to the entrance was at the north. At the north end of the north. The wall enclosed a layer of dark brown clayey structure there was a small patch of mid brown silt (context 861), which contained some charcoal sandy loam with a high concentration of charcoal and grains of hulled barley; it is interpreted as the (context 859). Inside and partly over the wall, there occupation deposit associated with the structure. A was a layer of mid brown sandy silt (context 856) radiocarbon date for this structure derives from a markedly less stony than the layers above. This

15 Site 8, phase I Site 8, phase II

868 868 866

788770 788770 866 868 864

Evaluation trench Evaluation trench

865 866 868 866 866 Limit of excavation 864 Limit of excavation Rock face Rock face

165900 165900

Site 8, phase III Site 8, phase IV

859 862 788770 788770 863 855 864 852 856

861 Evaluation trench Evaluation trench 857 863 862

Limit of excavation 852 Limit of excavation Rock face Rock face 852852

165900 165900

Wall stones

Illus 12 Plan of Site 8: (a) phase I: basal occupation deposit; (b) phase II: lower structure; (c) phase III: middle structure; (d) phase IV: upper structure.

16 deposit was thought to be decomposed turfs from the majority of the lithics are both interpreted as decom- walls of the structure. posed turfs that originated as collapsed wall The structure was dated from a sample of charcoal material. It is probable that the turfs used to taken from context 859. Again the calibrated radio- construct the walls were taken from the adjacent carbon date indicates a wide date range between ground and this contained prehistoric lithics. This AD 1660 and 1950, with the probability distributed suggests that the minor Bronze Age deposits encoun- over four peaks (Illus 11). However, as before, a 20th tered beneath the post-medieval shelters are only century date can be ruled out, suggesting the part of a more extensive occupation site that has structure dates to the 17th, 18th or 19th century AD. been disturbed by turf cutting.

Phase 5: Late features 6.2.5 Reuse and expansion over time

The remains of the three structures formed a The radiocarbon dates from this site show a similar platform up against the rock face. pattern of use as that of Site 6A. The early occupa- In the middle of this platform was an oval deposit tion phase was dated to the Middle Bronze Age, of dark greyish-brown sandy silt (context 857) filling roughly contemporary with the early phase at Site a shallow hollow. This was sealed by an area of 6A, followed by a gap of over 2000 years before packed stones (context 854) with a stone spread activity again resumed in the medieval period. The (context 853) to the north, forming the upper level of direct evidence for the Bronze Age phase was limited the platform. These stones were covered by turf and to two areas of charcoal-rich sediment with no associ- topsoil (context 851). The stone spread seems to be a ated structures and only one struck quartz flake. rough surface created up against the rock shelter but However, there is indirect evidence for more not associated with any surviving superstructure. substantial Bronze Age activity in the immediate This phase was given an upper age limit by finds vicinity through the lithics that were retrieved from retrieved from the top of the stones that comprised the turf walls of later structures. It is assumed that fragments of at least three glass vessels. These were the use of this site in the Bronze Age was determined all dated to the 20th century, not earlier than the by the shelter provided by the rock face, as it clearly 1920s (A Cox, pers comm). was in the later phases. However, it may also be noted that the sea-level displacement curve for Arisaig (Shennan et al. 1995) indicates that the site 6.2.4 Lithics by C Wickham-Jones was close to the sea shore at the time, overlooking a large shallow inlet of brackish water that later A small lithics assemblage (36 pieces) was retrieved evolved into the Mointeach Mhór. Proximity to the from Site 8. It largely comprised quartz flakes and bay with its natural resources – such as bird life, fish chunks and one quartz core. Some of these may be and shellfish – may have been another important natural, but it is more likely, given the presence of location factor although there is no direct evidence worked quartz, that they all either result from local for this in the archaeological record. knapping or have been brought on to the site as The lack of objects associated with the stack of late potentially useful. The quartz is a tabular, vein medieval and post-medieval structures supports the quartz that is presumably eroding out somewhere in suggestion that the site was not part of a permanent the locality. One large chunk has been used as a core settlement but is likely to have been used as a indicating that some quartz working took place on, temporary shelter, possibly by people herding live- or near to, the site. There are 18 quartz flakes, most stock. It may be contemporary with the two other of them broken. Breakage is probably a result of the small structures nearby, although these were not friable nature of the raw material and may have investigated and remain undated. The cultivation taken place during knapping. Many of the flakes are rigs in this area are assumed to be of 19th century of a good size and they would have been quite date, as they are elsewhere along the road line. suitable for use. In addition, two pieces of flints were retrieved. One is a broken bifacial point; the other a small broken 6.3 Rectangular turf structure: flake. The assemblage also contains one flake of Site 10 (NM 6583 8835) bloodstone that must have come from Rùm. Rùm bloodstone was used across a wide area of the west 6.3.1 Survey coast of Scotland from the Mesolithic onwards. The lithics assemblage was retrieved from con- The structure was located 420m to the south of Site 8 texts associated with all five phases, but mainly from on a low sand ridge at the western edge of Mointeach contexts 856 and 863 (Phases 2 and 4). It is not Mhór, probably a fossil beach ridge. It measured 3m thought that the lithics assemblage dates from the by 7.5m internally within turf banks 1.2–1.8m wide 16th to 20th centuries AD and must therefore have and up to 0.6m high. It was aligned north to south been re-deposited in all but the earliest (Bronze Age) with an entrance to the east. A square hollow 5m by sediments. The two sediments that produced the 5m located 5m to the north could possibly have been

17 created when the turfs were cut for the construction 6.4.3 Watching brief of the building. The building was located on the west side of a large rectangular area of improved ground, The area was closely watched during the topsoil 75m by 200m, depicted on the first edition OS map stripping prior to the construction of the road to surveyed in 1873. The CFA survey identified a record any early remains of the township not visible second structure just to the east, but this was not on the surface. A number of 19th-century glass and seen during the subsequent fieldwork. However, a ceramic finds were recovered from the topsoil and second rectangular turf structure 4m by 6m was two features were recorded: a large field bank seen recorded 70m to the south. during survey and an oval clearance cairn. The field bank comprised a single homogeneous deposit of earth with no evidence for a prolonged or complex 6.3.2 Evaluation history. Two small pits were also recorded, 0.3 and 0.5m in diameter and 0.1–0.2m deep. The larger and A trench was cut across the south wall of the shallower pit contained a primary fill with frag- building, exposing a well-defined turf wall with the ments of charcoal and burnt bone. The bone frag- individual turfs still visible. The wall was 1m wide ments were too small for identification and none of and 0.45m high. A possible occupation layer of white the pits contained any datable artefacts. The sand with dark patches was recorded inside the watching brief did not yield any evidence of an building, above the natural white sand. Six pottery earlier phase of the township pre-dating the 19th fragments were retrieved from the floor level of the century. building. These were identified as fragments of fine earthenware and a stoneware bottle. They were dated to the late 19th century or, more probably, the 6.5 Rectangular building: early 20th century. The site was interpreted as a late Site 26 (NM 6599 8693) 19th-century turf building and, as a result, no further investigation was carried out. 6.5.1 Survey This structure was located on the south-west side of a small knoll some 500m to the north of Arisaig 6.4 Township: Site 15 (centred on village. It comprised the footings of a rectangular NM 658 878) building aligned north-east to south-west that measured 3m by 7.7m internally, with a rectangular 6.4.1 Survey structure immediately to the north-east, possibly a lean-to structure built up against the east gable of This site comprised eight buildings with associated the building. The building was not recorded on any banks and cultivation remains situated to the west of maps so it was selected for evaluation to determine Kinloid overlooking Mointeach Mhór. It formed part its nature and date. of a now-abandoned settlement depicted on the first edition OS map and named Achraig. Three of the structures were depicted on the map as roofed 6.5.2 Evaluation buildings. The other five buildings were not mapped. Four buildings were selected for evaluation to test Two trenches 1m by 4m were excavated within the for the presence of earlier (pre-19th century) settle- site. One trench was located across the interior of the ment on the site. building, with the other trench cutting across the north-east wall and into the smaller structure. The walls were dry-stone random rubble construc- 6.4.2 Evaluation tion with a brown sand and gravel floor in the larger building and a roughly paved floor in the structure to Four buildings were selected for evaluation. In order the east. A couple of roof slate fragments were to increase the possibility of identifying earlier retrieved from the building but, unlike Site 15, the components of the township, three of the buildings structure produced no glass or ceramic fragments. were examples of those not depicted on the first The site was believed to be an isolated early edition OS map and were believed to pre-date the 19th-century building. It is not depicted on the first map. A trench was cut across the walls of each of the edition OS map and is therefore likely to have been structures to evaluate the date and nature of these demolished prior to 1870. sites. Two of the four structures were defined by turf and stone walls, the other two were built with dry-stone walls. The structures were interpreted as 6.6 Kerb cairn: Site 41 the remains of small heavily robbed enclosures and (NM 6634 8651) buildings. The finds assemblages from the sites were of early to mid 19th-century date. No evidence was The cairn was located during walkover survey along obtained for settlement pre-dating the 19th century. the road corridor carried out early in 2000 by the

18 25 CFA. It was located 20m east of the old A830 as it 786600 entered Arisaig from the east (Illus 13). The feature

22.5 appeared as a grassy mound 12m in diameter and 20 1m high. The nature of the monument was clarified A830 at the time through a small hand-excavated trench

17.5 that confirmed that the mound comprised small stones. The cairn was believed to be a funerary monument as the location made it less likely to be burnburn made from field clearance. This interpretation was

Kerb confirmed during an evaluation carried out by Headland Archaeology at the end of 2000, which identified the remains of a robbed out cist in the Excavated area centre of the structure and a boulder kerb on the 17.5

786500 north side of the cairn. As a result of the evaluation Cairn 15 results, a programme of full excavation was carried out during July and August 2001. Line of proposed road

12.5

To Arisaig 6.6.1 Excavation

15 N A trench measuring 30m by 30m was excavated, Building 0 50 m centred on the cairn. No features or deposits of ar- 166300 chaeological significance were identified in the area around the cairn. Bedrock was found to lie directly Illus 13 Contour map showing location of Site 41 below topsoil in many places and the cairn itself was constructed on top of an outcrop of bedrock (Illus 14). The first stage of construction comprised a 6m diameter kerb cairn (Illus 15). The kerb was built

Illus 14 Site 41: view of cairn after turf removal, from the south-west

19 Illus 15 Site 41: ring of kerb-stones, viewed from the north from large upright boulders supported on the inside Above the basal layer of cairn stones was a layer of by a layer of similarly large stones (context 4110). large voided stones (context 4102) within dark The remains of the robbed out cist (context 4105), red-brown silty sand. These were all located on the identified during the evaluation, were fully exposed interior of the kerb. Around the exterior of the kerb, and found to measure approximately 1m in length natural subsoil was overlain by a layer of stones and 0.45m in width, although this may not (context 4109) within dark brown silty sand, which represent the original size of the feature given that formed a concentric band around the kerb, roughly only two sides survived. No human remains or 2.5m in width. This layer abutted the boulder kerb artefacts were retrieved from within it. The cist was and had clearly been deposited after its con- supported by the basal layer of stones (4110) and struction. was thought to be contemporary with the erection of The whole cairn was overlain by a layer of smaller the kerb (Illus 16). stones (context 4101) in a matrix of light yellow- A small deposit of cremated human bone was iden- brown silty sand (Illus 18), forming a significant tified 3m to the north of the cist. It was recovered in mound measuring some 12m in diameter and at least an area measuring some 0.5m by 0.5m, within the 1m in height (Illus 19). Fifteen small sherds and basal layer of cairn stones (4110). The bones had not crumbs of prehistoric pottery were recovered from been interred in any cist-like structure but it is one location in the south-east quadrant. There are possible that they were deposited in an organic traces of comb-impressed decoration on several container that has totally decayed. The bone appears sherds and they have tentatively been identified as to have been deposited at the time of the cairn’s part of a Beaker (A MacSween, pers comm). In the construction although it is impossible to confirm that north-west quadrant of the cairn, a spread of stones were not moved to allow its insertion at a later charcoal (context 4103) was identified overlying date. Analysis of the remains suggested that they layer 4101. It measured some 1.0m by 0.6m and was were from one adult individual (D Henderson, pers only 7mm deep. Its origin was unclear and no small comm). A sample of the bones was dated by radio- finds were retrieved from the feature. A sample of carbon (SUERC-2451/GU-11904) and returned an the charcoal was dated by radiocarbon Early Bronze Age date between 2140 and 1910 BC (SUERC-2452/GU-11905) to the period between (Table 3 and Illus 17). 50 BC and AD 140 (Table 3 and Illus 17).

20 Illus 16 Site 41: plan of kerb and robbed out cist with basal layer of stones

21 SUERC-2451 (GU-11904) 3625–40BP

SUERC-2452 (GU-11905) 1950–40BP

4000CalBC 2000CalBC CalBC/CalAD 2000CalAD

Illus 17 Calibrated radiocarbon determinations from Site 41

Table 3 Radiocarbon determinations of samples from Site 41

Lab code Context Material Lab age BP dC13 1 sigma cal 2 sigma cal SUERC-2451 Find 2 Burnt human 3625 ± 40 –26.1‰ 2040–1910 BC 2140–2070 BC (12.7%) (GU-11904) bone 2050–1880 BC (82.7%) SUERC-2452 4103 Alnus 1950 ± 40 –26.8‰ AD 1–90 (58.5%) 50 BC to AD 140 (GU-11905) AD 100–30 (9.7%)

The centre of the cairn was dug into at some time the funerary cairn. The interpretation is after it had reached its final form. This left a depres- supported by the smaller size of the stones and the sion measuring 4m in diameter and 0.3m deep at the irregular extent of the deposit. present-day ground surface. Whenever this robbing The dating of the original construction of the cairn occurred, the result was partial destruction of the is dependent on the interpretation of the small cist and complete removal of its contents. cache of cremated human bone within the lowest level of the kerb cairn. If the bone is assumed to have been deposited during the original construc- 6.6.2 History and use tion (and there is no evidence to the contrary) and does not represent a token deposit of the curated This monument is a new example of a small remains of an ancestor, it can be used to date con- Bronze Age kerb cairn, a type of funerary struction to around 2000 BC.Thisisperfectly monument now familiar in the west of Scotland acceptable when compared with dates from similar (Lynch & Ritchie 1977). There are three other monuments (Lynch & Ritchie 1977). The presence unexcavated cairns in the Arisaig area that may be of a few small fragments of probable Beaker in a similar monuments (Illus 2). The nearest cairn lies stratigraphically late context on the edge of the some 200m to the south: it is a kerb cairn of similar cairn raises the possibility that the monument is in dimensions, 15m in diameter and 1m high, but it is fact significantly earlier. If the pottery represents badly robbed (NMRS no NM68NE 7). One kilometer the fragmentary remains of the grave goods from to the south lies a much larger cairn, 30m in the disturbed cist, it would suggest a construction diameter and 1.2m high (NMRS no NM68NE 6). date several centuries earlier in the third millen- The third cairn is situated 2.2 km to the south nium BC. In this case the cache of human bone would (NMRS no NM68SE 5); it measures only 5m in have been a secondary burial into the kerb cairn. It diameter and is 0.5m high so may not be a funerary is equally probable that the fragments of probable cairn like the others. Beaker are derived from another source – possibly The recorded structure and stratigraphy of the another burial nearby – and they were disturbed cairn allows for more than one interpretation of the and unintentionally deposited on the cairn during constructional sequence and history of the cairn. later field clearance. In this context, it is interesting The cist, although not central to the kerb, is appar- to note the Iron Age date obtained from the patch of ently a primary feature. The band of stones outside charcoal on the surface of the cairn. This suggests the kerb was clearly placed there after the kerb that the formation of the cairn, including the depo- was built but it does not totally obscure the kerb. It sition of field stones, was essentially complete by could therefore either be an original element of the this time and it may be used as evidence for cairn or an addition. The upper layer of smaller sustained cultivation of the land around the cairn stones is assumed to be an addition and may during later prehistory. The land surrounding the represent the gradual accumulation of field cairn remained in cultivation until the recent past clearance stones rather than a formal element of and is now improved pasture.

22 Evaluation Trench

4103

4101 Charcoal spread 4101

Outline of kerb 4107

4101 4101

786510

Area of disturbance N

Position of kerb stones beneath 4101

166340 0 3 m

Illus 18 Site 41: cairn after removal of topsoil

23 Profile 1

N S

Bedrock Profile 2 E W

Evaluation Trench (2000) Bedrock Evaluation trench

786520

786510 Profile 2 Cairn

166340 0 5 m Profile 1 Limit of excavation

Illus 19 Site 41: sections of the kerb cairn from (a) the north (Profile 1) and (b) the east (Profile 2)

24 7 Vegetation History

The archaeological programme was supported by colonized by alder carr at about 4950 BC. Alder palaeoenvironmental studies. A deep peat basin was declined at about 3200 BC to give way to a ling- identified during the initial archaeological survey of dominated heath; this was replaced by a bog myrtle the roadline at Allt Dail an Dubh-asaidh (Illus 2). A heath about 400 BC, which persists to the present peat core was extracted from this site and analysed day. The open marsh phase from 6100 to 4950 BC to provide a high-resolution record (both spatial and coincides with increased charcoal concentrations. temporal) of the vegetation of that area. A summary This is likely (at least in part) to be a result of the of the results and full consideration of their implica- temporary widening of the woodland canopy, also tions for the archaeology of Arisaig is included in this reflected in increased concentrations of airborne report. For a more detailed consideration of the data pollen, like pine. However, the presence of large (see Section 9 – Palynological Analysis) charcoal particles (>75mm), between 524 and 544cm in the core (dating roughly to 5900–5550 BC) is likely to be the product of burning close to the core site. The 7.1 Peat core from Allt Dail an cause of this burning cannot be determined from the Dubh-asaidh pollen core but may reflect activity by Mesolithic people in the immediate area. A 6.67m peat core was extracted and analysed for In the wider landscape, the vegetation was birch– pollen and charcoal at intervals of 4cm, giving a total hazel–oak woodland with some pine and elm of 160 samples. Radiocarbon dating of pollen zones throughout the early prehistoric period. High per- throughout the core has indicated a date range from centages for tree pollen, alongside continued but low 8000 BC to the present day for the sediments. levels of disturbance, suggest some kind of woodland Analysis of the core at a resolution of approximately use between 3200 and 400 BC. Woodland cover is 50 years has illustrated a continuous Holocene vege- maintained until 3200 BC when it starts to decline tation sequence from open water prior to 8000 BC up gradually, associated with elevated levels of char- to the present-day mix of acid heath and birch– coal. Large charcoal particles (>75mm) probably hazel–oak woodland. reflect burning events close to the pollen core site The pollen derives from both local vegetation (that between 236 and 256cm (roughly 1350–1520 BC). is, within the small basin that the peat was accumu- From 3200 BC there is evidence for low levels of vege- lating) and from extra-local and regional vegetation. tation disturbance relating to human activity with The pollen sequence can be divided into six zones ruderal pollen types (like ribwort plantain, nettle that primarily reflect changes in the local vegetation and goosefoot) indicating disturbance, consistently as the original lochan gradually filled with peat recorded. The sequence is interpreted as reflecting (summarized in Table 4). Changes in the extra-local extensive grazing impacts which, particularly from and regional pollen are much less clear-cut and must about 600 BC, maintained a regionally open vegeta- be isolated from the more obvious local changes in tion with discrete pockets of woodland. After 600 BC, order to identify change in the wider landscape. while the decline in woodland pollen continues, Within the basin, the original open-water lochan along with sustained curves for ruderal pollen types, was progressively colonized by willow carr which, while charcoal frequencies decline. This suggests after an open marsh phase from 6100 BC with sustained but low levels of disturbance with human grasses, sedges and some fen species, was in turn activity throughout the area but with less impact

Table 4 Pollen zones, identifying main local and extra-local and regional vegetation types Zone Depth Calibrated date Local (on site) Extra-local and regional (cm) (approx) vegetation vegetation 1F 0–176 400 BC to present day Sedge–ling–bog myrtle heath Birch–hazel–oak woodland 1E 176–368 3200–400 BC Ling heath Birch–hazel–oak woodland 1D 368–480 4950–3200 BC Alder carr Birch–hazel–oak woodland with elm and pine 1C 480–544 6100–4950 BC Open marsh with sedges and Birch–hazel–oak woodland with grasses elm and pine 1B 544–640 8000–6100 BC Willow carr with sedges and Birch–hazel–elm woodland with grasses pine 1A 640–base (666) Before 8000 BC Open water Willow carr

25 from burning activity. The sustained disturbance Dubh-asaidh, pollen and diatom records were used that causes continued declines in tree pollen and an to look at sea-level change in the area (Shennan et al. increase in open vegetation is likely to be livestock 1995). Entire pollen sequences were not analysed grazing. and most sections from Mointeach Mhór dated to 1250 BC to just after 780 BC (Shennan et al. 1995). This period is characterized by a sharp decline in 7.2 Other evidence for vegetation woodland pollen at 780 BC with a slight increase in history of the Arisaig area ribwort plantain, indicating levels of disturbance with grasses and ling. This, with similar evidence The results from Allt Dail an Dubh-asaidh provide from Allt Dail an Dubh-asaidh, appears to reflect detailed data on only one small part of the Arisaig regional woodland decline between about 3200 and landscape. Fortunately, there are pollen diagrams 650 BC. Pollen analyses from Loch nan Eala were also available from four other locations in or very close to undertaken for information on sea-level changes. Arisaig that can be used to assess variations in This core has ten stratigraphically distinct sections landscape history across this area (Illus 2). Of the that were dated to between 10,500 and 7050 BC, and four pollen diagrams available, all of them are dated, at c 7150, 5500, 2520, 2140, 1930 and 1730 BC. From although some are incomplete sequences. There is a about 2500 BC at this site, ruderal pollens, including complete Holocene sequence from Lochan Doilead in ribwort plantain, are present, accompanied by lower North Morar (Williams 1977), partial diagrams from values for woodland pollen types. The core, again, the Mointeach Mhór and Loch nan Eala in Arisaig seems to reflect the wider regional decline in (Shennan et al. 1994; Shennan et al. 1995) and a later woodland from about 2500 BC, although interpreta- Holocene sequence from Polnish, east of Borrodale tion of the data in terms of vegetation history is (Clarke 2000). The changes in tree pollen percent- difficult given the short and separate sections ages from Lochan Doilead, Allt Dail an Dubh-asaidh analysed. This is particularly unfortunate as Loch and Polnish are summarized in Illus 20. nan Eala lies within the potential ‘core’ settlement Williams demonstrates an elm decline at 3550 BC, area of Arisaig and could contain evidence for a marking the start of woodland decline in North different landscape history over the past 5000 years. Morar, which is steady and progressive from here to Further to the east, a short core from a woodland the top of the diagram (Williams 1977). Later, at hollow has been dated and analysed (Clarke 2000) 2150 BC, the start of continuous curves for ruderal and provides a vegetation record from about 1850 BC pollen types reflects continuous, although low, levels to the present. Decline in woodland pollen is much of disturbance accompanied by more marked less apparent in this core and this may reflect its woodland decline. Woodland recovers at about 500 BC location in an area of steep slopes that is wooded at and fluctuates from the Iron Age to the present day, present. High total tree pollen values are main- although there is an overall sustained decline. It is tained for 2900 years from 1850 BC but the proportion only at about AD 600 that heath pollen types start to of birch and oak pollen varies, with an increase in increase, most notably with the start of the increase birch possibly reflecting exploitation of oak during in ling (Calluna vulgaris) pollen. Overall, the history this period. Charcoal increases from about AD 1020 to of gradual but progressive woodland decline at the top of the core and is likely to be reflecting Lochen Doilead from 3550 BC is very similar to that increased human activity and burning. Increased obtained from Allt Dail an Dubh-asaidh. grazing at the top of the sequence has also been Just to the south of the pollen site at Allt Dail an inferred (Clarke 2000).

26 8 Occupation and Use of a Marginal Landscape

8.1 Core and periphery in Arisaig that ensure people keep returning to it despite other options being available. The land surrounding Sites In the introduction to this paper it was suggested 3–6 must be a candidate for a traditional shieling that the existing archaeological record for Arisaig ground where these social ties operated. The supports the identification of a core area of perma- repeated reuse of a few preferred sites has a major nent settlement from as early as the Bronze Age, influence on archaeological visibility. Contrary to surrounded by peripheral areas of less-intensively what might be expected in a landscape that has expe- exploited land. The data collected in the course of the rienced low-intensity, extensive human land-use, A830 investigations support this model, with archaeological sites are likely to be disturbed and relevant evidence coming both from site-specific obscured as sites are repeatedly reused. Effectively, archaeological studies and the landscape-scale palaeo- they are isolated islands of highly intensive land-use environmental study. with only the most recent structures visible on the The site-specific data are limited in terms of the surface. However, both the lack of any evidence in number of sites but are clear-cut. The proposed core the archaeological record for Iron Age activity and area yielded a Bronze Age kerb cairn, the third or the apparent change in woodland cover and distur- fourth recorded example from this area, neatly rein- bance levels in the pollen record perhaps reflect a forcing the restricted distribution of this site type. cultural discontinuity at this time. Human activity is The later history of the kerb cairn includes modifica- evident, but seems to have moved from the favoured tion and disturbance resulting from agriculture and Bronze Age sites in the peripheral areas, and was other activities, reflecting the dynamic nature of the having an irregular impact on the woodland vegeta- landscape within an area of permanent settlement. tion. Activities in the core area seem to have been The pollen evidence reflects higher levels of distur- modifying Bronze Age land-use practices, moving bance and impact on the woodland vegetation from from burial to cultivation. Later in medieval times, the Bronze Age (Illus 20) with patchy regeneration the peripheral Bronze Age sites were again adopted. during the Iron Age and beyond. This is likely to be The pollen record reflects gradually intensifying reflecting a change in land use in both the core and land-use, resulting in an overall loss of woodland peripheral areas. with irregular phases of regeneration during this The peripheral area yielded examples of shieling time. huts and shelters (with dates spanning the medieval and post-medieval periods) and minor, but poorly understood, sites of Bronze Age date. It also con- 8.2 Site-based and landscape-scale tained long open spaces with no structural evidence evidence for human impact for human activity (if the 19th-century sites are put to one side for the moment) and this emphasizes the The concentration of the site-based archaeological highly selective use of the landscape. There is a evidence into a few preferred locations in the periph- strong contrast between empty areas and selected eral area might give the impression that human sites to which people have repeatedly returned. At impact was also restricted in its extent. But this is to Sites 3–6, there is a relatively sheltered and well- confuse human impact with visible archaeological drained hill slope with running water that has been ‘sites’; the key human impact in the peripheral areas occupied over at least the 1000 years of the medieval has been the extensive modification of vegetation and post-medieval periods as well as in the Bronze and, indirectly, of soils caused by the felling of trees Age. At Site 8, there is a structure that was re-built and grazing of livestock. The rare archaeological on at least three occasions over a period of a few sites are therefore of little value in gauging the centuries, each time taking advantage of a vertical nature and degree of human impact. In the present rock face to provide both shelter and one side of a project there is no excavated evidence for human lean-to type of shelter. Again, there is evidence that activity in the peripheral area before the Middle this precise spot was also occupied in the Bronze Age. Bronze Age or for 2000 years before the medieval Repeated reuse of the same location presumably period, yet the pollen record from Allt Dail an reflects the natural advantages of the site: for Dubh-asaidh documented progressive removal of example water supply, drainage or shelter, which woodland from the late Neolithic period through to are all restricted commodities in this landscape. the present day without any significant reversals in Most of the peripheral area is simply too rocky, too this process. In fact, in the Arisaig area, no marked exposed or too wet to be occupied. Continuity of use decline in woodland has ever taken place. This is may also reflect the influence of tradition where, unusual in Scotland and either reflects continuously once established, a site acquires strong social ties low levels of occupation or use, or sustained

27 Polnish 100

40 % tree pollen tree % 30

0 9000BC 8000BC 7000BC 6000BC 5000BC 4000BC 3000BC 2000BC 1000BC 0 1000AD 2000AD

Lochan Doilead 100

% tree pollen tree % 30

0 9000BC 8000BC 7000BC 6000BC 5000BC 4000BC 3000BC 2000BC 1000BC 0 1000AD 2000AD

Allt Dail an Dudh-asaidh

100

% tree pollen tree % 30

0 9000BC 8000BC 7000BC 6000BC 5000BC 4000BC 3000BC 2000BC 1000BC 0 1000AD 2000AD

Illus 20 Changes in tree pollen percentages from Polnish, Lochan Doilead and Allt Dail an Dubh-asaidh

28 management of the woodland resource. This gradual decline in woodland, being punctuated with periods of removal and regeneration, could relate to the apparent change in land-use reflected in the archaeological record by a change from visible Bronze Age sites to currently invisible Iron Age ones. It may be argued that the information obtained from the excavation of Sites 6 and 8 has contributed very little to an understanding of the nature of human activities in the past. For deposits where radiocarbon dates could be obtained, we can conclude that something happened at a particular time in the past, but the nature of that ‘something’ is impossible to define from the available evidence. The selective locations of these sites tells us about the ways in which people have ‘read’ and occupied their landscape but it is the peat-derived pollen and charcoal record that offers coherent data on the nature of human land-use and its changing impact through time. So, whilst it is satisfying as archaeologists to have detected the stacked sequences of occupation deposits in the shieling sites, we should recognize the limitations and high cost of excavation and be willing to invest sufficient resources in palaeoenvironmental approaches.

8.3 Evolution of the Arisaig landscape

The ultimate aim of this project has been to understand how the landscape of Arisaig evolved through time. The geographical limitations of the project were stated at the start of this paper and these can only be partially overcome through the use of data from other studies. Nevertheless, we believe that there is now sufficient data with which to propose a model of landscape evolution, if only to encourage critical testing by future researchers of its more spec- ulative aspects. Three stages have been identified in the evolution of the landscape although they should probably be viewed as points along a continuum rather than persistent stages separated by relatively brief periods of transition (Illus 21a–c).

8.3.1 Stage 1: Early prehistory (7500– 3550 BC), Illus 21a

We know from the published excavations at Kinloch on Rùm (Wickham-Jones 1990) and the more-recent work of the Scotland’s First Settlers Project that the west coast of Scotland was populated by human communities from as early as 7500 BC. However, it

Illus 21 (left) Stages of landscape evolution near Arisaig: (a) early prehistory: 7500–3550 BC; (b) later prehistory: 3550 BC to AD 500; (c) medieval and later period: AD 500–1800 (from Ordnance Survey maps © Crown copyright)

29 appears that there was no appreciable impact on the likely to have taken place. Woodland continued to be natural landscape of Arisaig throughout the Meso- used for grazing. lithic and early Neolithic periods up until approximately 3550 BC, and Illus 21a shows a complete cover of woodland without attempting to plot the variation 8.3.3 Stage 3: Medieval and later period that must have existed in its species composition (AD 500–1800), Illus 21c because of topographical variation. Human communities in this period exploited natural woodland In this third stage of landscape evolution, three resources without significantly modifying them and distinct types of land-use can be defined. The core there is no evidence for a conventional Mesolithic– settlement areas, with significant patches of culti- Neolithic transition from mobile hunter-gatherer to vated land, have expanded along the sandy soils of sedentary farmer. The pollen evidence indicates that the coast to include settlements such as Traigh and the woodlands probably were used for grazing, if not Glenancross (both settlements recorded in early as sources of wood, although there is no direct rentals of Arisaig), perhaps reflecting increasing evidence for woodland management in the archaeo- human populations. Woodland has become more- logical or palynological record. In Arisaig, this may or-less restricted to defined areas (such as the pollen reflect the fact that the areas that were later site at Polnish) where it is protected from grazing preferred as small core areas of settled farmland and managed for sustained production of timber. were largely underwater during this period, which Remaining areas are essentially open heath and bog coincides with the mid-Holocene sea-level maximum utilized for extensive livestock grazing. (Shennan et al. 1995; Shennan et al. 2002). The coastline shown in Illus 21a probably existed between 5900 and 3200 BC. Farming has subse- 8.3.4 Postscript: The recent past (AD 1800– quently developed on patches of raised beach and 1900) coastal shell sand that are freely draining, less rocky and less acidic than the rest of the landscape, which One final point should be made in this brief land- appears never to have been suitable for cultivation. scape history. The archaeological record in Arisaig is So, early in the Neolithic period, agriculture may not dominated by the remains of settlement and agricul- have been a viable option, although pastoral farming ture dating to the 19th century. Evaluation of probably was and herds of animals were grazing in selected sites during this project yielded artefact and around the woodlands. assemblages that suggest much of this settlement dates from a short period in the early to mid-19th century, and OS mapping from the 1870s shows that 8.3.2 Stage 2: Later prehistory (3550 BC to buildings were already in ruins and cultivation rigs AD 500), Illus 21b abandoned by this date. The detailed history of this period can be explored in contemporary documents Progressive differentiation of the landscape is and this major task lay outside the resources and detectable from 3550 BC and it is assumed that the ambitions of the present project. Suffice to say that core farming area of Arisaig began to develop from this brief period (perhaps no more than 50 years) this time, exploiting the limited areas of freely appears as a major discontinuity in an archaeolog- draining raised beach at the head of Loch nan Ceall ical record that otherwise emphasizes continuity of although direct archaeological evidence is lacking land-use and landscape development. Settlements until late in the third millennium BC. Other areas of were built where none was before; land was culti- land that emerged as relative sea level fell, for vated for the first and probably only time. The example the Mointeach Mhór, were too wet and origins of this anomalous period no doubt lie in the rapidly developed into extensive salt marshes, catastrophic social and economic upheaval that through alder carr to ling-dominated bog. overtook the western Highlands in the aftermath of There is no archaeological evidence as yet that the the Highland Clearances: an event that finally broke coastal shell sands were exploited. There has been no the continuity of several millennia of landscape attempt to date these deposits and it is possible that evolution in Arisaig. the on-shore movement of sand did not occur until Given the magnitude and high visibility of this late in prehistory and therefore the development of event, as reflected in archaeological sites, it is what are now favourable soils for agriculture was important that we note the failure of it to register in also late. Pollen data show that woodland remained the available pollen records for Arisaig. There are a extensive throughout later prehistory. Fluctuations number of possible explanations for this apparent in woodland cover during the Late Bronze Age and anomaly. Pollen from cereal and potato crops does into the Iron Age reflect periods of loss and regenera- not disperse far so the short-lived fields were tion. Regeneration in the Iron Age appears patchy probably not close enough to the pollen sites reported and irregular, suggesting land-use changed with the here to register. Similarly, the construction of new resulting impact on the fragmenting woodlands. It is settlements would not in itself register in the wider assumed that significant clearings were restricted to vegetation record. It might be assumed that a signifi- the core settlement area, where agriculture is more cant increase in population would lead to impacts on

30 the surrounding landscape through increased well-documented episode of major social and demand for fuel (peat and wood) and for livestock economic change does not register in the pollen grazing rights, but this is only an assumption. The record we must remain cautious in our use of this absence of this effect may simply reflect the short- source of information about past landscapes. Pollen lived nature of the 19th-century events but it is also only provides a proxy record of past human activity possible to interpret this as evidence that the pollen and our interpretation of the data must always record is complacent in this regard. If a recent and reflect this fact.

31 9 Palynological Analyses at Allt Dail an Dubh-asaidh

9.1 Introduction and Summary Pollen samples were prepared following standard procedures (Moore et al. 1991); Hydrofluoric acid (HF) This section presents the results of pollen analysis of was used to remove silica at the base of the profile. deep peat deposit at Allt Dail an Dubh-asaidh (NGR: Marker spores were added to allow the calculation of NM 6714 9054) on the line of the re-aligned A830 pollen concentrations (Stockmarr 1972). Identifica- road, between Arisaig and Mallaig. tions were achieved using standard keys and a The 6.67m pollen core was analysed at a sample comprehensive reference collection; pollen taxonomy resolution of 4cm, giving a total of 160 samples. follows the Cataolgue of Pollen Types (Bennett 1995). Radiocarbon dating of pollen zones throughout the English names for plants, pollen and spores are used core has indicated a date range from 8900 BP to the where possible throughout the report, although the present day. Analysis of the pollen core at a resolu- pollen diagrams use Latin names. Where possible, tion of approximately 50 years has illustrated a full bog myrtle (Myrica gale) and hazel type (Corylus Holocene sequence from open water prior to 8900 avellana type) pollen have been separated. However, years BP. identifications cannot be made consistently (Edwards This open water was colonized by willow carr, 1981) and hence the bog myrtle curve in this diagram which after an open marsh phase with grasses, represents a minimum, while the hazel curve in Zone sedges and some fen species, was in turn colonized by 1F is likely to be enhanced through the inclusion of alder carr at about 6000 years BP. Alder declined at undifferentiated Myrica pollen. Counts to at least 300 about 4500 years BP to give way to a ling-dominated land pollen grains were made to achieve statistically heath locally with birch–hazel–oak woodland in the reliable counts, although at some levels this total was wider area, with pine and elm. From this time, there not achieved because of high accumulation rates, is continuous evidence for low levels of disturbance particularly in Zones 1A and 1C. Each grain was also relating to human activity. In addition, the extra- assigned to one of five preservation classes (well- local and regional birch–hazel–oak woodland shows preserved, corroded, degraded, crumpled and a gradual decline, with, at first, high frequencies of broken), according to the dominant state of preserva- charcoal. With a local change to bog myrtle heath, tion (cf Cushing 1967, Tipping 1987). Routine counts charcoal frequencies decline; woodland pollen of microscopic charcoal were made and were calcu- continues to decrease and ruderal pollen types, indi- lated as concentrations. Pollen percentages were cating disturbance, are consistently recorded. From calculated on the basis of total land pollen (TLP). 4500 BP, the sequence is interpreted as reflecting Other types were calculated as a percentage of TLP extensive grazing impacts which, particularly from plus group. about 2500 years BP, maintained a regionally open The resolution adopted here at 4cm gives an vegetation, with ruderal herbs and a continuing but approximate resolution of 50 years, which in Zones gradual decline in woodland. This is proposed as an 1E and 1F where human activity is reflected in the extensive, yet continuous, model of human use of the vegetation and archaeological record represents west coast of Scotland from late Neolithic times to approximately two generations in human terms. the present day.

9.3 Radiocarbon dates 9.2 Methodology Six pollen zones have been identified, primarily An eigelkampf corer was used to take 3-cm diameter reflecting changes in vegetation within the basin cores of sediments along two transects (see Illus 22) from open water through marsh to alder carr and from across the Allt Dail an Dubh-asaidh basin at then heath. Five radiocarbon dates taken at the zone 70m OD. In total 12 cores were sampled to assess boundaries have demonstrated that a full Holocene basin profile (Illus 23, Illus 24). At the deepest point sequence is present with no evidence for truncation (marked 4* on Illus 22–24), a Russian corer was used or major hiatuses. Accumulation rates have varied to retrieve a 6-cm diameter core for palaeo- between 0.5 and 1.0cm per decade, giving a high environmental analyses with a sediment block for temporal resolution (Table 5). the top 20cm. Stratigraphic descriptions of the sediments were carried out in the laboratory on cleaned sediment surfaces. Descriptions use the 9.4 Results conventions of Troels Smith in Table 5 (Troels Smith 1955). Colours were described using international Results are presented in a percentage pollen dia- Munsell colour charts. gram (Illus 25), and the main pollen, fern and moss

32 Table 5 Pollen zones with radiocarbon dates and accumulation rates Pollen Depth Summary Radiocarbon Accumu- Sample zone (cm) description date (from base lation rate resolution of zone), BP (cm/year) (years/4cm) 1F 176–0 Sedge–ling heath with birch–hazel–oak woodland 2410 ± 75 0.07 63 1E 368–176 Ling heath with birch–hazel–oak woodland 4520 ± 60 0.09 44 1D 480–368 Alder carr with Birch–hazel–oak woodland and 6070 ± 65 0.07 57 elm and pine 1C 544–480 Open marsh with birch–hazel–oak woodland plus 7245 ± 65 0.05 80 pine and elm 1B 640–544 Willow carr with birch–hazel–oak woodland with 8900 ± 100 0.06 67 pine 1A 667–640 Open water with willow carr – – –

Illus 22 Sketch map showing core locations

33 Illus 23 Long traverse basin depth profiles

Illus 24 Short traverse basin profiles spore curves are also presented as concentration 9.5 Interpretation of pollen data curves in Illus 26. These additional curves remove the relative influence of the percentage data on 9.5.1 Zone 1A: Base to c 8900 BP curves and are useful in discerning actual changes in the pollen curves. Illus 27 shows pollen preservation In the clay and silty clays of this zone, pollen concen- data. Table 6 details the stratigraphic description of trations are low and there are some remnants of a the pollen core. late glacial-type flora, including crowberry (Empe- The pollen diagram has been divided into six trum nigrum) and meadow rue (Thalictrum), zones, based on dominant pollen types. Table 4 (see indicative of tundra conditions. However, locally, the Section 7.1) summarizes the main characteristics of pollen spectra are dominated by aquatic plants, the pollen zones and indicates the dominant local growing in slow-flowing/non-moving water. This (within the basin) and extra-local and regional vege- community was surrounded by willow carr with tation (on the basin sides and further afield). grasses and sedges. There is no archaeological

34 Table 6 Stratigraphic descriptions of pollen core (Troels Smith notation) Depth (cm) Munsell colour Troels Smith description Notes 0–67 Very dark brown Well-humified turfa peat Grass and sedge roots 10YR 2/2 Th2 Sh2 64–140 Black Well-humified peat with herbacea Fragments of sedge 5YR 2.5/1 fragments Sh3 Th1 Dh+ 140–159 Black Well-humified peat with herbacea and Grass and sedge fragments with 10YR 2/2 woody fragments woody stems (ling) @ 150 Sh2 Th1 Dh1 Dl+ 159–227 Very dark brown Very well-humified peat with herbacea Woody stems (ling) @ 182 & 194 5YR 2/2 and woody fragments Large wood piece @ 209–210 Sh4 Th+ Tl+ Dl+ 227–324 Black Well-humified peat with herbacea and Fragments of grass and sedge stems, 5YR 2.5/1 woody fragments also ling Sh3 Th1 Tl+ Dg+ Dl+ 324–326 Well-humified detritus peat Sh2 Dl2 326–338 Black Well-humified peat with herbacea and Fragments of grass and sedge stems, 5YR 2.5/1 woody fragments also ling Sh3 Th1 Tl+ Dg+ Dl+ 338–404 Black Very well-humified peat 5YR 2.5/1 Sh4 Th+ Dl+ 404–408 Black Well-humified peat with herbacea and Large wood pieces (diam > 2cm) 10YR 2/1 woody fragments Sh2 Dg1 Dl1 Th+ Dh+ 408–465 Black Very well-humified peat Grass and sedge stem fragments 5YR 2.5/1 Sh4 Th+ Dl+ 465–471 Black Well-humified peat 10YR 2/1 Sh3 Th1 Dh+ 471–651 Black Well-humified peat with silt 5YR 2.5/1 Sh4 Th+ Tl+ Dh+ As+ 651–657 Dark grey Clay with organic and silty elements Mica present 10YR 4/1 Ag3 Sh1 Th+ Dh+ As+ 657–662 Greyish-brown Silty clay Mica present 10YR 5/2 As2 Ag2 Sh+ Dh+ 662–667 Very dark greyish-brown ClayAg Mica present. Occasional herbaceous 10YR 3/2 4 Sh+ Dh+ stem remains

evidence for any human activity at this time and the From 624cm, near the base of this zone, oak pollen spectra do not indicate any evidence of human (Quercus) and elm (Ulmus) start to increase. Beyond activity. the local carr, a mixed woodland with birch, hazel, elm, oak and pine (Pinus diploxlon type) occurred. The presence of elm, in particular, in the extra-local 9.5.2 Zone 1B: c 8900–7245 BP and regional woodland is shown in the pollen concentrations in Zone 1B. Locally, willow carr dominates and has colonized the There are some indications of very low levels of open water in Zone 1A. As a component of the local disturbance in the vegetation reflected by the occa- vegetation, willow (Salix) is a low pollen producer sional presence of ruderal herbs but it is difficult to (Bradshaw 1981) and its local presence is only shown relate these entirely to human activity. Such low in the concentration curves. Birch (Betula) and hazel levels of disturbance could equally be the result of (Corylus type) were also present locally in the natural disturbance in the birch–hazel woodland. canopy. The understory was composed of sedges (Cyperaceae), grasses (Poaceae), ling (Calluna vulgaris), Sphagnum mosses with some fen types, 9.5.3 Zone 1C: c 7245–6070 BP including meadowsweet (Filipendula ulmaria). Sphagnum is likely to have been restricted to The pollen diagrams both show an open marsh with discrete pockets where ground conditions were more sedges and grasses and some fen herb species in Zone acidic. Otherwise the vegetation indicates a neutral 1C. The increases in pine pollen and microscopic to slightly base environment. charcoal may reflect an increased recruitment area

35 Illus 25a Allt Dail an Dubh-asaidh: percentage pollen diagrams for trees and shrubs

36 Illus 25b Allt Dail an Dubh-asaidh: percentage pollen diagrams for herbs

37 Illus 25c Allt Dail an Dubh-asaidh: percentage pollen diagrams for acquatics

38 Illus 26a Allt Dail an Dubh-asaidh: pollen concentration curves for trees and shrubs

39 Illus 26b Allt Dail an Dubh-asaidh: pollen concentration curves for herbs, ferns and mosses

40 Illus 27 Allt Dail an Dubh-asaidh: pollen condition classes

resulting from the opening up of the willow carr. This and, given the very low levels of evidence in the opening of the canopy may also account for the pollen record, is likely to have been of low impact and increase in bracken spores, although bracken pro- dispersed across the wider pollen catchment. duces spores on an irregular basis and the apparent The sediment accumulation rates are slow through- increase may have been caused by other factors out Zone 1C (see Table 5) and this is reflected in the (Page 1982). Pine, as a component of the extra-local low pollen concentrations throughout this zone (Illus and regional woodland, is unlikely to have increased 26). This is also suggested by the increase in corroded in occurrence in Zone 1C. Pine is notoriously over- pollen grains, which may also indicate fluctuating represented in pollen diagrams, particularly those water tables (Illus 27). If this is the case, the higher from open sites (Bennett 1984; Fossit 1994) and this concentrations of elm and pine pollen grains, which probably accounts for the apparent increase in pine are both relatively robust pollen types, are also illus- in this zone. Both pine and elm pollen concentrations trating poorer preservation conditions as well as a remain approximately constant from Zone 1B, wider catchment area. suggesting no real increase in trees in Zone 1C. It is likely that the vegetation of Zone 1C was The extra-local and regional areas would have subjected to low levels of disturbance, possibly with been characterized by woodland dominated by birch some influence from human activity. However, given and hazel with oak, elm and pine. Concentrations of the larger catchment of Zone 1C, this activity could birch and hazel have declined in Zone 1C, which may have taken place throughout the wider area at the be reflecting some low levels of woodland use and low levels of impact recorded in the pollen record. clearance by human populations. Of the herb types recorded, there is little evidence for higher levels of disturbance in open areas. 9.5.4 Zone 1D: c 6070–4520 BP The charcoal curve indicates sustained level of burning in the area throughout Zone 1C, which could Zone 1D is dominated by the rise in alder (Alnus be reflecting some human activity in the wider area, glutinosa) pollen, with ferns and herbs (Illus 25). The although again the larger catchment area, resulting alder rise at other pollen sites in the Arisaig area is a from the decrease in local tree canopy, is likely to be clearly recognizable feature of pollen diagrams, inflating charcoal concentrations.. This activity although dates for its inception vary. The start of the would be taking place in a late Mesolithic context alder rise at the end of Zone 1C, dated here to about

41 6000 BP, is much earlier than the alder rise dated at bance include nettle and goosefoot. Their sustained Mointeach Mhór, slightly further south (Shennan et curves indicate sustained levels of disturbance. al. 1995). This is reflecting local conditions and may Zone 1E straddles the mid-Neolithic to the end of reflect the local topography slightly above the Moin- the Bronze Age. The locally open vegetation teach Mhór; the latter would have been subject to the dominated by ling with grasses and heather reflects maximum Holocene sea-levels at this time. At a wider pollen catchment area and indicators of low Lochan Doilead, the alder rise has also been dated to levels of human activity in the wider area are more c 6000 years BP (Williams 1977). evident. Zone 1E is also characterized by the start of This alder carr will have developed locally, with a a gradual decline in tree pollen, illustrated in the dense canopy and the herbs represented would have summary curves on Illus 25. All the tree curves show been elements of the understory. These included progressive declines throughout Zone 1E, reflecting Ranunculus acris type, which includes meadow steady continuous decrease in tree canopy. The slow buttercup, medowsweet, bedstraws (Galium type) decline of elm throughout Zone 1E is comparable to and some ferns including lady fern (Athyrium felix- the decline in other tree pollen concentrations, which femina) and broad buckler fern (Dryopteris dilatata). suggests low levels of woodland clearance in discrete Extra-locally, woodland would have been character- areas, but not necessarily favouring elm, one of the ized by birch, hazel and oak woodland with elm and theories for the widespread elm decline. Humans, as pine. Zone 1D covers the early Neolithic period and, factors in woodland decline are implicated by the rise with the dense local alder carr, any human activity in ruderal herb pollen and charcoal, but the low outside the woodland is difficult to see because the levels of evidence reflects an extensive use of the filtering effect of the alder carr is significant. area. This evidence for human activity in Bronze Age There is no evidence in this zone. For the ubiqui- times is supported by archaeological evidence for tous decline in elm pollen at about 5100 BP across Late Bronze Age sites in the Arisaig area. north-west Europe (Whittington & Edwards 1997). The zone is also marked by microscopic charcoal The elm pollen concentrations in Zone 1D are consis- throughout, which after the sharp increase halfway tently high and gradually decrease through Zone 1E through the zone at 296cm, continues throughout along with other tree pollen. Low levels of woodland the zone. Again this is likely to be reflecting human clearance dispersed throughout the area would be activity. Extra-locally the woodland remains charac- compatible with a gradual decrease in all woodland terized by birch and hazel, with lower levels of oak, tree pollen, where any vector for elm decline, elm and pine. whether felling or dutch elm disease, probably the most likely causes (Rackham 1990), would be dispersed throughout a large area. The lack of a visible 9.5.6 Zone 1F: c 2410–present elm decline at about 5100 BP is likely to be reflecting low levels of human activity, in the form of felling Zone 1F is marked by the local change from a and opening up the woodland and was dispersed ling-dominated heath to one with increased grasses, widely enough across the area to prevent a marked sedges and bog myrtle (Myrica gale). Despite the fact decline in elm pollen. that the curve for bog myrtle is a minimum, because of the difficulties in distinguishing consistently between pollen from hazel and bog myrtle (Edwards 9.5.5 Zone 1E: c 4520–2410 BP 1981), there is nonetheless a definite change in the pollen record from ling to a bog myrtle heath. This The start of Zone 1E is defined by the decline in the type of succession has been associated with changes alder pollen. This is replaced locally by ling with in climate, burning or grazing (Rodwell 1991). Micro- heather (Erica tetralix and E. cinerea) and grasses. scopic charcoal is present at low levels throughout The zone starts with a short transitory sedge phase. the zone. So unless the trigger was the consistently This mirrors the vegetation sequence illustrated higher charcoal levels throughout Zone 1E, over a further south at Mointeach Mhór (Shennan et al. 2000-year period, then burning is unlikely as a major 1995), following the decrease of alder, dated later at cause. The date at the start of Zone 1F (2410 BP) that site to 2565 ± 45 years BP. Again this later date coincides with the end of the Bronze Age and a reflects local site conditions at Mointeach Mhór. It is generally accepted change in climate to wetter condi- now widely accepted that the rise and fall of alder tions across Scotland (eg Dubois & Ferguson 1985; pollen in the early to mid-Holocene is driven by local Gear & Huntley 1991). This could have triggered the topographic and environmental factors rather than a change from woodland to heath with increased continuous spread from source areas (Whittington & grasses, sedge and bog myrtle. The decline in Edwards 1997). woodland could also be related to increased grazing Although there are only low levels of indicators for throughout Zone 1F, although this is difficult to human activity throughout the diagram in Zone 1E distinguish directly in the pollen record (see below). there is a continuous evidence for ribwort plantain It is more likely to be a combined change to a wetter (Plantago lanceolata), which is associated with climate alongside increased grazing regimens, disturbed habitats (Godwin 1975; Grime et al. 1992). reflected in the woodland decline but with low levels Other ruderal herbs recorded that reflect distur- of burning.

42 Around the local sedge and bog myrtle heath, birch phase allowed the colonization of the local area by and hazel woodland with oak is found. Although the alder carr. Towards the end of this phase, acidifica- concentrations of hazel and birch seem high for an tion seems to occur as alder dies away and ling takes apparently open environment, these values are over the basin. The increase in bog myrtle in Zone 1F likely to be reflecting the local woodland to the south is likely to relate to the continuing influence of water and east. In addition, research by Bunting indicates inflow, helping it compete with ling. that these levels of arboreal pollen are consistent Changes in the extra-local and regional woodland with an open environment on the west coast of are harder to interpret, although there are clearly Scotland with woodland in local pockets (Bunting fluctuations in canopy composition and extent. The 2002). The increase in pine pollen towards the top of changes from willow- to alder-dominated carr are Zone 1F (from 12cm) is likely to be reflecting modern likely to reflect natural successional changes, as plantations. above. In the wider landscape, the birch–hazel–oak Throughout Zone 1F there are clear indications of woodland remains constant in Zone 1D, and its sustained disturbance in the vegetation, although gradual decline starting in Zone 1E and continuing again at consistently low levels. Curves for ruderal through Zone 1F is likely to relate more to human herbs, including ribwort plantain, nettle, sorrel and activity and possibly woodland decline as a result of goosefoot are sustained throughout. The pollen increased grazing. catchment area is gradually widened throughout Zone 1F because of the decline in woodland cover, although charcoal frequencies remain low, suggest- 9.8 Human activity ing that burning was not a major activity. It is more likely that the sustained ruderal pollen curves and There has been little archaeological evidence the gradual decline in tree pollen reflect increasing recorded for human activity in the area around Allt grazing levels throughout the region. Dail an Dubh-asaidh until the 19th and 20th centuries and there is similarly little evidence for 9.6 Interpretation of charcoal data human activity affecting the vegetation, until Zone 1F, after about 2400 years BP. Most of the main changes in vegetation at the pollen site can be The charcoal curve in these diagrams (Illus 25, Illus accounted for through natural succession. 26) shows increased burning in Zones 1C and 1E. However, human activity reflected in the pollen From 7200 to 6000 BP in Zone 1C, this is likely to record is at a consistently low level throughout the reflect an increased pollen catchment area, following diagram from the Mesolithic period to the present the opening up of the willow carr characterizing Zone day. Woodland cover at a local and extra-local and 1B. There is very little pollen evidence for human regional scale is maintained to the start of Zone 1E at activity throughout Zone 1C, which is consistent about 4500 BP. From this time, woodland declines with the early dates, although the increased charcoal gradually and is likely to be reflecting human reflects a higher frequency of burning events than activity throughout the area at sites local to and could be assumed to be natural. further away from the pollen core site. This In Zone 1E, from 4500 to 2400 BP, burning frequen- sustained human activity at low levels throughout cies increase, with a decline in tree pollen and the area is supported by the curves for ruderal pollen indicators of disturbance in the form of ruderal types like ribwort plantain, nettle and goosefoot. herbs. The burning in this zone is likely to be a result In Zone 1F, while the decline in woodland pollen of human activity, although again at low levels and continues along with sustained curves for ruderal dispersed throughout the area. pollen types, charcoal frequencies decline. This suggests sustained but low levels of disturbance with 9.7 Natural vegetation succession human activity throughout the area but with less impact from burning activity. The sustained disturbance that causes continued declines in tree pollen Natural processes of succession account for most of and an increase in open vegetation is likely to be the local vegetation changes recorded at Allt Dail an increased grazing. Dubh-asaidh. The colonization of open water by willow carr would have occurred with sedimentation of the basin. Willow carr occupies generally neutral to basic conditions and the apparently odd inclusion 9.9 Conclusions of Sphagnum in this local community is likely to reflect pockets where slight acidification can occur, The vegetation record at Allt Dail an Dubh-asaidh, perhaps at the basin edge (cf Salix cinerea–Betula from the detailed pollen record described here, pubescens–Phragmites australis woodland, Rodwell indicates low levels of disturbance, related to human 1991). However, the generally neutral to basic activity from about 4500 years BP onwards. Before nature of the area is maintained and Sphagnum this time, any human activity would have caused disappears with the development of an open marsh even lower levels of disturbance that, although characterized by sedges and grasses. This open possible to discern in the pollen diagram, are likely to

43 have been masked by natural vegetation succession very gradual decline in tree pollen. It is also likely at the pollen site. that grazing played an increasingly important role in From 4500 BP, the pollen record includes low but the vegetation. Although there are very few direct continuous levels of human disturbance. This would indicators for grazing in the pollen record, the impact be consistent with small discrete areas of human of increased grazing is reflected in the suppression of activity that were used in an extensive pattern, on a tree pollen production and the maintenance of open temporary basis. This small-scale activity, in the form vegetation types, particularly through Zone 1F. From of woodland clearance, with some burning, was not 2500 BP onwards, therefore, the vegetation was necessarily in the same location and it is more likely disturbed on a continuous basis, probably grazing. that the activity moved from one small area to This could have been either through transhumance another through time. This would produce the low but practice throughout the area or through maintained sustained levels of ruderal herb pollen along with a grazing at a network of small sites.

44 10 Acknowledgements

Archaeological investigations of the A830 road re- Analysis of the excavated material has benefited alignment at Arisaig spanned several years and from the work of a number of specialist consultants involved many individuals and organizations. The and we are grateful for contributions from Adrian Cox entire archaeological project was funded by the (post-medieval artefacts), Caroline Wickham Jones Scottish Executive Transport Group and managed on (lithics) and Mhairi Hastie (plant remains). Julie its behalf by Patrick Ashmore of Historic Scotland. Mitchell (University of Edinburgh) is thanked for Initial stages of the project were undertaken by the undertaking the pollen preparations and Paula Centre for Field Archaeology, University of Milburn for help with pollen analysis. Jane Bunting, Edinburgh, and we are grateful to staff of CFA for the Althea Davies and Eileen Tisdall all contributed to ready provision of information through the later stages discussions on interpretations of the pollen sequence of the fieldwork and preparation of this publication. and we are grateful to Ciara Clark for allowing us to Fieldwork was facilitated by the various contractors refer to her unpublished pollen data from Polnish. and project managers working on the road construc- The illustrations for this report were drawn by Laura tion project and we particularly wish to acknowledge Speed, Tom Small and Lisa Wilson of Headland the assistance of Bob Spence (Highland Council, Archaeology. Resident Engineer), Mike Horton (Ritchies), David Stevenson and Billy Johnston (Barr Construction).

45 11 References

Bennett, K 1984 ‘The post glacial history of Pinus Page, C N 1982 The Ferns of Britain and Ireland. sylvestris in the British Isles’, Quat Sci Rev 3, Cambridge. 133–155. Rackham, O 1990 Trees and Woodland in the Bennett, K 1995 Catalogue of Pollen Types: Mid British Landscape. Phoenix. and Late Quaternary Sediments. Cambridge. Rees, A 1996 ‘Polnish chapel to Loch Nan Uamh Bradshaw, R 1981 ‘Modern pollen representation viaduct’, Disc Excav Scot 1996, 58. factors for woods in south-east England’, Ritchie, A, Ritchie, G, Whittington G, Soulsby, J Journal of Ecology 69, 45–70. 1974 ‘A prehistoric field-boundary from the Branigan, K, Foster, P 1995 Barra: Archaeological Black Crofts, North Connel, Argyll’, Glasgow Research on Ben Tangaval. SEARCH, Vol 1. Archaeol J 3, 66–70. Sheffield. Rixon, D 2002 Arisaig and Morar: A History. East Bunting, M J 2002 ‘Detecting woodland remnants Linton. in cultural landscapes: modern pollen Rodwell, J 1991a British Plant Communities: Vol deposition around small woodlands in 1. Woodlands and Scrub. Cambridge. northwest Scotland’, The Holocene 12, 291–301. Rodwell, J 1991b British Plant Communities, Vol Clarke, C 2000 A830–Polnish Bridge to Loch nan 2. Mires and Heaths. Cambridge. Uamh Improvement: Palynological Analysis Shennan, I, Innes, J, Long, A, Zong, Y 1994 ‘Late from a Peat Core Near Arnabol Burn. CFS Devensian and Holocene relative sea-level report no 528. Edinburgh. changes at Loch nan Eala, near Arisaig, Cushing, E 1967 ‘Evidence for differential pollen northwest Scotland’, J Quat Sci 9, 261–283. preservation in late Quaternary sediments in Shennan, I, Innes, J, Long, A, Zong, Y 1995 ‘Late Minnesota’, Rev Palaeobotany Palynology 4, Devensian and Holocene relative sea-level 87–101. changes in northwestern Scotland: new data to Davies A L, Tipping, R 2005 ‘Sensing small-scale test existing models’, Quat Intl 26, 97–123. human activity in the palaeoecological record: Shennan, I, Peltier, W R, Drummond, R, Horton B fine spatial resolution pollen analyses from P 2002 ‘Global to local scale parameters West Glen Affric, northern Scotland’, The determining relative sea level; changes and the Holocene 14. post-glacial isostatic adjustment of Great Dubois, A D, Ferguson, D K 1985 ‘The climatic Britain’, Quat Sci Rev 21, 397–408. history of pine in the Cairngorms, based on Stockmarr, J 1972 ‘Tablets with spores used in radiocarbon dates and stable isotope analysis, absolute pollen analysis’, Pollen et Spores 13, with an account of the events leading up to its 615–621. colonization’, Rev Palaeobotany Palynology 46, Tipping, R 1987 ‘The origins of corroded pollen 55–80. grains at five early post-glacial pollen sites in Edwards, K 1981 ‘The separation of Corylus and western Scotland’, Rev Palaeobotany Myrica pollen in modern and fossil samples’, Palynology 53, 151–161. Pollen et Spores 23, 205–218. Troels Smith, J 1955 ‘Characterisation of Fossitt, J 1994 ‘Modern pollen rain in the unconsolidated sediments’, Danmarks northwest of the British Isles’, The Holocene 4, Geologiske Undersolgelse series IV 3, 38–73. 365–376. Whittington, G, Edwards, K 1997 ‘Vegetation Gear, A, Huntley, B 1991 ‘Rapid changes in the change’ in Edwards, K, Ralston, I (eds) range limits of Scots pine 4000 years ago’, Scotland: Environment & Archaeology, Science 251, 544–547. 8000 BC – AD 1000. London. Grime, J P, Hodgson, J G, Hunt, R 1992 The Wickham-Jones, C 1990 Rhum: Mesolithic and Abridged Comparative Plant Ecology. London. Later Sites at Kinloch. Excavations 1984–86. Godwin, H 1975 The History of the British Flora. Society of Antiquaries of Scotland Monograph Cambridge. Series Number 7. Edinburgh. Lynch, F, Ritchie J N G 1977 ‘Kerb-cairns’, Proc Williams, W 1977 The Flandrian Vegetational Soc Antiq Scot (1974–75) 106, 30–33. History of the Isle of Skye and the Morar Moore, PD, Webb JA, Collinson, M 1991 Pollen Peninsula. Unpublished PhD thesis, University Analysis, 2nd Edn. London. of Cambridge.